markat1/mistral-7binstruct-summary-100s

README.md

@@ -20,7 +20,12 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the generator dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.5987
+- eval_loss: 2.0198
+- eval_runtime: 34.7344
+- eval_samples_per_second: 2.706
+- eval_steps_per_second: 0.345
+- epoch: 0.06
+- step: 15
 
 ## Model description
 
@@ -39,7 +44,7 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 4e-05
+- learning_rate: 2e-05
 - train_batch_size: 1
 - eval_batch_size: 8
 - seed: 42
@@ -47,45 +52,9 @@ The following hyperparameters were used during training:
 - total_train_batch_size: 2
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
-- lr_scheduler_warmup_steps: 0.1
+- lr_scheduler_warmup_steps: 0.03
 - training_steps: 150
 
-### Training results
-
-| Training Loss | Epoch | Step | Validation Loss |
-|:-------------:|:-----:|:----:|:---------------:|
-| No log        | 0.02  | 5    | 2.4606          |
-| 2.4794        | 0.04  | 10   | 2.2401          |
-| 2.4794        | 0.06  | 15   | 2.0792          |
-| 2.1147        | 0.08  | 20   | 1.9539          |
-| 2.1147        | 0.1   | 25   | 1.8629          |
-| 1.888         | 0.12  | 30   | 1.8053          |
-| 1.888         | 0.14  | 35   | 1.7636          |
-| 1.7809        | 0.16  | 40   | 1.7317          |
-| 1.7809        | 0.18  | 45   | 1.7081          |
-| 1.7043        | 0.2   | 50   | 1.6918          |
-| 1.7043        | 0.22  | 55   | 1.6788          |
-| 1.6923        | 0.25  | 60   | 1.6678          |
-| 1.6923        | 0.27  | 65   | 1.6579          |
-| 1.6453        | 0.29  | 70   | 1.6491          |
-| 1.6453        | 0.31  | 75   | 1.6416          |
-| 1.5791        | 0.33  | 80   | 1.6347          |
-| 1.5791        | 0.35  | 85   | 1.6279          |
-| 1.6144        | 0.37  | 90   | 1.6219          |
-| 1.6144        | 0.39  | 95   | 1.6168          |
-| 1.6472        | 0.41  | 100  | 1.6124          |
-| 1.6472        | 0.43  | 105  | 1.6086          |
-| 1.5904        | 0.45  | 110  | 1.6056          |
-| 1.5904        | 0.47  | 115  | 1.6033          |
-| 1.5891        | 0.49  | 120  | 1.6016          |
-| 1.5891        | 0.51  | 125  | 1.6004          |
-| 1.5882        | 0.53  | 130  | 1.5996          |
-| 1.5882        | 0.55  | 135  | 1.5991          |
-| 1.6018        | 0.57  | 140  | 1.5988          |
-| 1.6018        | 0.59  | 145  | 1.5987          |
-| 1.581         | 0.61  | 150  | 1.5987          |
-
-
 ### Framework versions
 
 - PEFT 0.9.0

adapter_config.json

@@ -9,13 +9,13 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 16,
+  "lora_alpha": 64,
   "lora_dropout": 0.1,
   "megatron_config": null,
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 8,
+  "r": 32,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cd820b60b88b9e51818cf5f5d266c456d82a971952320ae8bb106a5b9d2537e3
-size 13648432
+oid sha256:c60b8684e4fc174c64ee65169402174c73475b89d27b094a13d5b595591bc30c
+size 54543184

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:80ded19b3f3184c49ad15ef2837df275e08cac6ca7769bfaa777728dd7d25d8b
+oid sha256:e328bed096c7f01176f4ccaa21df959a73c03b0f481e67f80ac001bf2d078c94
 size 4920